6636
J. Phys. Chem. B 2008, 112, 6636–6645
Inhomogeneous NMR Line Shape as a Probe of Microscopic Organization of Bicontinuous Cubic Phases Konstantin I. Momot,*,† K. Takegoshi,‡ Philip W. Kuchel,§ and Timothy J. Larkin§ School of Physical and Chemical Sciences, Queensland UniVersity of Technology, GPO Box 2434, Brisbane, Qld 4001, Australia, Department of Chemistry, Kyoto UniVersity, Kyoto, Japan, and School of Molecular and Microbial Biosciences, UniVersity of Sydney, Sydney, NSW 2006, Australia ReceiVed: January 22, 2008; ReVised Manuscript ReceiVed: February 27, 2008
NMR line shapes of the lipid and aqueous species in bicontinuous cubic phase (BCP) samples prepared by centrifugation are inhomogeneously broadened. The broadening of the lipid peaks is removed by magic-angle spinning (MAS). In this work, we studied the mechanism of this broadening using 1H and 13C NMR spectroscopy of a myverol/water BCP. It is demonstrated that the inhomogeneity possesses an intrinsic contribution that is independent of instrumental or setup factors and can be attributed to the microscopic organization of the BCP bilayer. A mechanism of the inhomogeneous broadening is proposed, which involves a spatially nonuniform diamagnetically induced magnetic field determined by the mesoscopic structure and the diamagnetic susceptibilities of the two BCP domains. The proposed mechanism does not require that molecular reorientation of the lipid be slow for the inhomogeneous broadening to survive. We discuss how this inhomogeneous broadening can be employed as a probe of compositional uniformity and microscopic organization of BCP samples. Introduction BCP are thermodynamically stable lipid/water liquid crystalline systems characterized by a mesoscopic periodic structure possessing medium- to long-range cubic periodicity.1,2 In the static limit, the geometry of the lipid bilayer can be approximated by an infinite periodic minimal surface (IPMS); two congruent but nonintersecting channel systems are occupied by aqueous medium. A schematic diagram of a BCP of the Ia3d spatial symmetry is shown in Figure 1. BCPs have found uses in controlled-release drug delivery3–7 and crystallization of membrane proteins8,9 and have provided an interesting in vitro model for studying the effects of curvature on the properties of the lipid bilayer.10,11 These applications have presented the need for the understanding of functional properties and rational design of BCP-based and similar materials, motivating studies of their microstructure and molecular mobility.12–18 NMR lines of the lipid and aqueous species in BCP samples are inhomogeneously broadened.19 Several potential causes of this broadening have been proposed. MAS has been shown to reduce or eliminate the broadening, resulting in improved NMR spectral resolution as compared to the static samples.19 MAS diffusion measurements of related lyotropic systems have also provided insights into the diffusional behavior of the lipid.20 Recently, a new line-narrowing technique, consisting of loading BCP samples into thin-walled glass capillaries, has been discovered.21 In this work, we focus on investigating the mechanism of the inhomogeneous broadening in BCP samples. This is done with a view to using NMR line shape inhomogeneity as a probe of the microscopic organization of BCPs, as has previously been done in other materials.22–25 We present a number of observations demonstrating the intrinsic nature of the inhomogeneous line broadening in BCPs, most importantly, * To whom correspondence should be addressed. Fax: +61 7 3138 1521. E-mail:
[email protected]. † Queensland University of Technology. ‡ Kyoto University. § University of Sydney.
the differential MAS line narrowing between the water and the lipid species. We used the myverol/H2O system, which is well-characterized and forms an Ia3d BCP above 23 °C near the composition point 70% lipid:30% H2O (w/w).26 Nonspinning 1H NMR spectra of this system exhibited distinctly inhomogeneous line shapes. We demonstrate that at least some of this inhomogeneous broadening is independent of the experimental setup and conditions and is intrinsic to the lyotropic lipid/water cubic phase. We discuss a number of candidate mechanisms of the line broadening observed and propose one in which the survival of the inhomogeneous broadening does not require molecular reorientation to be slow. This mechanism involves a nonuniforminduced magnetic field in a system consisting of two domains with different diamagnetic susceptibilities (“lipid” and “water”). The proposed mechanism is supported by the results of numerical simulations of the induced magnetic field in model lipid/water systems. Methods Sample Preparation. BCP samples were prepared from water and myverol (CAS 85586-30-7). Myverol is a mixture of monoacylglycerol lipids, in which 1-monoolein (CAS 2549672-4) is the major component. It is frequently used as an inexpensive alternative to monoolein because of the similarity of their lipid/water phase diagrams.26 Three BCP samples were examined as follows: sample 1, myverol:H2O 70:30%; sample 2, 75.9:24.1%; and sample 3, 75.2:24.8% (all w/w). Cubic phase samples were prepared by combining the required quantities of lipid and saline in a centrifuge tube. The samples were homogenized by repeated cycles of centrifugation at 12000g, manual stirring, and incubation at 25 °C over a period of 2-3 days. Following preparation, the samples were stored at 25 °C to avoid the coexistence of the lamellar crystalline phase (Lc).26 NMR measurements were performed within 3 days of sample preparation. The chemicals used were obtained from the following sources: Myverol 18-99 was from Quest International
10.1021/jp8006415 CCC: $40.75 2008 American Chemical Society Published on Web 05/06/2008
NMR Line Shapes in BCPs
J. Phys. Chem. B, Vol. 112, No. 21, 2008 6637 1H
Figure 1. Schematic drawing of the Ia3d BCP. The average center of the bilayer follows the gyroid minimal surface.
(Zwijndrecht, The Netherlands); PBS (osmolality, 265 ( 1 mmol kg-1; 0.01 M phosphate buffer, 0.0027 M KCl, and 0.137 M NaCl, pH 7.4, at 25 °C) was prepared from PBS concentrate (Sigma-Aldrich, United States). Myverol was dried by rotary evaporation at 52 °C followed by freeze drying and was stored under argon.16 Other reagents were used as received. High-Resolution Solution Spectra. Spectra of myverol solutions in CD3OD were recorded at B0 ) 9.4 T using a Bruker DRX-400 NMR spectrometer equipped with a 5 mm TXI probe. The spectrometer has been described previously,27,28 and standard NMR procedures were used. The spectra were recorded from static 5 mm samples. Static BCP Spectra. BCP samples 1 and 2 were loaded either in a 5 mm Shigemi NMR tube susceptibility-matched to D2O or in a 8 mm flat-bottom NMR tube inserted in an outer 10 mm tube filled with CCl4.16 In both cases, the length of the sample was constrained to 8-9 mm using custom-made Delrin or Teflon inserts, and the cylindrical symmetry axis was parallel to B0. Constraining of the length enabled the containment of the sample within the region of homogeneous B0 and, in diffusion measurements, constant field gradient.27 Coaxial orientation of the cylindrical samples relative to B0 minimized shimming inhomogeneities. The homogeneity of B0 was adjusted typically to 3-5 Hz using the residual HDO signal from a 5 mm bulk D2O sample; shimming was optimized again following the placement of the BCP sample. The BCP samples were packed until no significant air pockets could be seen. NMR measurements were made with at B0 ) 9.4 T using the same spectrometer as for the solution spectra. 1H spectra were recorded using a diffusion probe equipped with either a 10 mm
or a 5 mm 1H/19F resonator. 13C spectra were recorded at natural abundance and without 1H decoupling using the same diffusion probe with a 10 mm 13C resonator. The pulsed field gradients (used in diffusion measurements) have a maximum strength 10 T m-1, and the coils are actively shielded. The sample temperature (25 °C in all measurements) was calibrated using a capillary containing methanol.29–31 NOESY spectra were acquired using a gradient-selected, phase-sensitive pulse sequence32 with a bipolar gradient pulse applied during the mixing time. Diffusion and T1 measurements were made as described previously.16 Exponential apodization with 3 Hz line broadening was used for most spectra; this enabled the enhancement of the signal-to-noise ratio without significantly affecting the spectral line widths. Spectral line widths at half-height were determined by using curvilinear least-squares fitting of Lorentzian functions onto the observed line shapes. The line widths of overlapping peaks were determined using simultaneous leastsquares fitting of the overlapping peaks. Further experimental details are given in the literature.16,27 MAS NMR Spectroscopy. MAS NMR measurements (the spinning angle 54.7° relative to B0) were made using a Tecmag Apollo spectrometer operating at the 1H frequency 300 MHz (B0 ) 7.05 T). The spectrometer was equipped with a dualchannel Chemagnetics MAS probe. MAS rotors were manufactured of zirconia and had outer and inner diameters of 7 and 6 mm, respectively. The spinning speed of the sample was set using a Chemagnetics tachometer/MAS speed controller and measured precisely using the frequency separation between spinning sidebands in the NMR spectrum. Two Teflon spacers of 10 mm length were placed inside the rotor to constrain the sample. The Teflon spacers and the rotor cap were glued to the rotor using epoxy resin to prevent the ejection of the sample at large spinning speeds. The magic angle was adjusted using a KBr powder sample following the standard procedure.33 T1 relaxation times were measured using a conventional inversion-recovery pulse sequence. Simulations of the Magnetic Field Maps in BCPs. The locally induced static magnetic field in the lipid and the aqueous domains of an Im3m BCP unit cell was simulated numerically using the surface-current technique.34 The model used in the simulations assumed that the center of the lipid bilayer followed a Schwartz P surface geometry; the bilayer was characterized by a constant, specified thickness, and the magnetic susceptibilities of the aqueous and the bilayer domains were -9.0 and -5.0 ppm, respectively. The simulation code was written in Fortran and based on the code previously used for calculating static magnetic field maps.35,36 The surface-current integration was carried out over a single Im3m unit cell with its center of symmetry placed at the origin. Because the integration was performed over a single unit cell, the distribution of the induced magnetic field B(r) in the aqueous domain was computed only for the grid points located within the “inner” aqueous subdomain (the subdomain containing the origin). The integration procedure was adapted to noncylindrical symmetry by modifying the differential surface element used for surface-current integration:35
B(r) ) µ0 4π
∑ ∫∫ [ i
Si
]
B0(∆χ)i ei × (r - r ′ ) µ0 |r - r ′ |3
|Ri| R (φ, z) dφ dz (1) (Ri · ei) i
where Si is the surface of the i-th lipid-water interface; the index i refers to the inner or the outer surface of the bilayer (which is given a finite thickness expressed as a fraction of the
6638 J. Phys. Chem. B, Vol. 112, No. 21, 2008
Momot et al.
length of the side of the Im3m unit cell); vector r′ spans the respective lipid-water interface; vector r spans the volume of the Im3m unit cell; (∆χ)i is the magnetic susceptibility difference at interface i; ei is a unit vector tangential to the surface of interface i and perpendicular to B0; Ri(φ, z) is the vector normal to the z-axis and connecting the z-axis with the point r′; [Ri, φ, z] are the cylindrical coordinates of vector r′; and |Ri(φ, z)| ) Ri(φ, z). Typical integration parameters were as follows: 20 × 20 × 20 grid points, 4000 surface integration steps in each of the vertical (z) and the angular (φ) dimensions, and 0.02 unit cell sizes as the minimal allowed distance between a grid point and a lipid-water interface. The simulations were performed on a desktop PC (Pentium 4 CPU, 3.2 GHz, 1 GB RAM). Results Static Samples. 1H NMR spectra from well-shimmed, nonspinning BCP samples with the cylindrical symmetry axis of the sample parallel to B0 were characterized by line shapes 70-100 Hz (B0 ) 9.4 T). A representative 1H NMR spectrum of sample 2 is shown in Figure 2a. The line widths at halfheight of the H2O peak and five well-resolved lipid peaks are shown in Table 1. The different lipid line widths exhibited by the two samples are due to their slightly different compositions; this effect is discussed in the next section. In CD3OD solution (5% myverol by weight), the respective lipid peaks exhibited a well-resolved multiplet structure (Figure 2b); the line widths of the multiplet components were approximately 0.01 ppm. The line widths of representative, well-resolved 13C NMR peaks are shown in Table 2. The T1 values of H2O and selected lipid peaks were measured from static sample 1 at B0 ) 9.4 T. T1 values of the lipid were also measured in CD3OD solution. The results of all T1 measurements are shown in Table 3. The T2 values of H2O in static sample 1 were measured using the CPMG experiment with 180° pulse separation times τ ) 4, 6, 8, and 10 ms. The T2 values were 185 ( 5 ms and were independent of τ. The values of the diffusion coefficients of H2O and lipid in static sample 1 were 4.2 × 10-10 and 1.2 × 10-11 m2 s-1, respectively. Two-dimensional 1H NOESY spectra of sample 1 were recorded with the mixing times τ ) 200 and 600 ms. The spectrum recorded at τ ) 600 ms is shown in Figure 3a. A distinctive feature of these spectra was the elongated shape of the NOE cross-peaks between lipid peaks. Density maps of a selected well-resolved cross-peak, together with the half-height contours, are shown in Figure 3b (600 ms) and c (200 ms). A NOESY spectrum of 5% (w/w) myverol in CD3OD was also recorded; no significant NOEs between lipid peaks were detected in this sample. MAS Measurements. The line widths of one-dimensional 1H and 13C spectra were measured under MAS conditions at a series of spinning speeds ranging from 5 to 2030 Hz. At slow spinning speeds (
